DEEP TRANSFER LEARNING FOR RETINAL IMAGE ANALYSIS IN DIABETES PREDICTION: A SYSTEMATIC REVIEW

Main Article Content

Kokila A
R. Shankar
S. Duraisamy

Abstract

Diabetes is a chronic metabolic disorder that affects millions globally, leading to severe complications such as diabetic retinopathy (DR), which can result in vision impairment or blindness. This systematic study examines the function of deep transfer learning (TL) in the early detection of diabetes using retinal image analysis. It comprehensively examines many pre-trained convolutional neural networks (CNNs) and hybrid models employed in diabetic prediction, highlighting their architectures, datasets, performance metrics, and clinical applicability. Furthermore, this paper discusses the challenges of retinal image classification, including dataset imbalances, interpretability, and computational complexity. The study also identifies key future research directions to improve the robustness, generalizability, and clinical deployment of AI-driven diabetic prediction systems. The findings suggest deep TL (DTL) is a promising tool for noninvasive, automated, and scalable diabetes screening using retinal images.

Downloads

Download data is not yet available.

Article Details

Section
Articles